OBJECTIVE: Prediction of the effect of a home physiotherapy intervention on the basis of four clinical characteristics of patients with advanced Parkinson's disease. DESIGN: A repeated measures design comparing six weeks without treatment with six weeks of physiotherapy and a follow-up of 12 weeks. SUBJECTS: Persons with Parkinson's disease without dementia and suffering from considerable functional disability. INTERVENTION: Community physiotherapists treated patients in the home situation three times a week teaching cueing and conscious movement control for walking and carrying out transfers in and out of beds and chairs. MAIN OUTCOME MEASURES: Mental status, disease severity, age and mood were included as predictor variables. A new functional scale developed as part of a previous study was used as the dependent variable administered in both the hospital and the home to determine whether the outcome generalized from the learning to a different environment. RESULTS: Only disease severity was a negative predictor of treatment outcome at home. In the hospital setting none of the factors predicted the immediate effect of treatment but cognitive ability and age were determinants of whether the treatment effects were maintained in the long term. CONCLUSIONS: Using cueing and cognitive strategies benefited younger and older patients with Parkinson's disease alike. However, the findings indicate targeting of treatment at patients with milder disease severity and providing follow-up treatment for older and cognitively less able patients.